19 Power System Reliability

The electric power industry began in the late 1800s as a component of the electric lighting industry. At this time, lighting was the only application for electricity, and homes had other methods of illumination if the electricity supply was interrupted. Electricity was essentially a luxury item and reliability was not an issue.

As electricity became more common, new applications began to appear. Examples include electric motors, electric heating, irons, and phonographs. People began to grow accustomed to these new electric appliances, and their need for reliable electricity increased. This trend culminated with the invention of the radio. No nonelectrical appliance could perform the same function as a radio. If a person wanted to listen to the airwaves, electricity was required. As radio sales exploded in the 1920s, people found that reliable electricity was a necessity. By the late 1930s, electricity was regarded as a basic utility (Philipson and Willis, 1999).

As electric utilities expanded and increased their transmission voltage levels, they found that they could improve reliability by interconnecting their system to neighboring utilities. This allowed connected utilities to “borrow” electricity in case of an emergency. Unfortunately, a problem on one utility’s system could now cause problems to other utilities. This fact was made publicly evident on November 9, 1965. On this day, a major blackout left cities in the northeastern United States and parts of Ontario without power for several hours. Homes and businesses had become so dependent on electricity that this blackout was crippling. Action was needed to help prevent such occurrences from happening in the future.

19.1 NERC Regions

The North American Electric Reliability Corporation (NERC) was formed in 1968 as a response to the 1965 blackout. By this time, reliability assessment was already a mature field and was being applied to many types of engineered systems (Billinton and Allan, 1988; Ramakumar, 1993). NERC’s mission is to promote the reliability of the North America’s bulk power system (generation and transmission). It reviews past events; monitors compliance with policies, standards, principles, and guides; and assesses future reliability for various growth and operational scenarios. NERC provides planning recommendations and operating guidelines, but has no formal authority over electric utilities.

Since most of the transmission infrastructure in the United States and Canada is interconnected, bulk power reliability must look at systems larger than a single utility. The territory covered by NERC is far too large to study and manage as a whole, and is divided into 10 regions. These NERC regions are as follows: East Central Area Reliability Coordination Agreement (ECAR), Electric Reliability Council of Texas (ERCOT), Florida Reliability Coordinating Council (FRCC), Mid-Atlantic Area Council (MAAC), Mid-Atlantic Interconnected Network (MAIN), Mid-Continent Area Power Pool (MAPP), Northeast Power Coordinating Council (NPCC), Southeastern Electric Reliability Council (SERC), Southwest Power Pool (SPP), and the Western Systems Coordinating Council (WSCC). The geographic territories assigned to the 10 NERC regions are shown in Figure 19.1.

Even though there are 10 NERC regions, there are only four major transmission grids in the United States and Canada: the area associated with the WSCC, the area associated with the ERCOT, Quebec, and the eastern United States. These are usually referred to as the Western Interconnection, the ERCOT Interconnection, the Quebec Interconnection, and the Eastern Interconnection. Each of these grids is highly interconnected within their boundaries, but only has weak connections to the other grids. The geographic territories associated with these four interconnections are shown in Figure 19.1.

NERC looks at two aspects of bulk power system reliability: system adequacy and system security. A system must have enough capacity to supply power to its customers (adequacy), and it must be able to continue supplying power to its customers if some unforeseen event disturbs the system (security). Each of these two aspects of reliability is further discussed in the following.

FIGURE 19.1 NERC regions.

19.2 System Adequacy Assessment

System adequacy is defined as the ability of a system to supply all of the power demanded by its customers (Billinton and Allan, 1988). Three conditions must be met to ensure system adequacy. First, its available generation capacity must be greater than the demanded load plus system losses. Second, it must be able to transport this power to its customers without overloading any equipment. Third, it must serve its loads within acceptable voltage levels.

System adequacy assessment is probabilistic in nature (Allan et al., 1994; Schilling et al., 1989). Each generator has a probability of being available, P_A, a probability of being available with a reduced capacity, P_R, and a probability of being unavailable, P_U. This allows the probability of all generator state combinations to be computed. A simple two-generator example is shown in Table 19.1. There are nine possible generator state combinations, and the probability of being in a particular combination is the product of the individual generator state probabilities. In general, if there are n generators and x possible states for each generator, then the number of possible generator state combinations is

$G e n e r a t o r s t a t e c o m b i n a t i o n s = x^{n}$ $G e n e r a t o r s t a t e c o m b i n a t i o n s = x^{n}$

(19.1)

In addition to generator state combinations, loading behavior must be known. Information is found by looking at historical load bus demand in recent years. For the best accuracy, 8760 h peak demand curves are used for each load bus. These correspond to hourly peak loads for a typical year. To reduce computational and data requirements, it is usually acceptable to reduce each set of 8760-h load curves to three weekly load curves (168 h each). These correspond to typical weekly load patterns for winter conditions, spring/autumn conditions, and summer conditions. Weekly load curves can be scaled up or down to represent temperatures that are above or below normal. Sample weekly load curves for a winter peaking load bus are shown in Figure 19.2.

To perform an adequacy assessment, each generation state combination is compared with all hourly loading conditions. For each combination of generation and loading, a power flow is performed. If the available generation cannot supply the loads or if any constraints are violated, the system is inadequate and certain loads must be shed. After all generation/load combinations are examined, the adequacy assessment is complete.

An adequacy assessment produces the following information for each load bus: (1) the combinations of generation and loading that result in load interruptions and (2) the probability of being in each of these inadequate state combinations. From this information, it is simple to compute the expected number of interruptions for each load bus, the expected number of interruption minutes for each load bus, and the expected amount of unserved energy for each load bus. These load bus results can then be aggregated to produce the following system indices:

TABLE 19.1 Generator State Probabilities

Generator State
Generator 1	Generator 2	Probability
Available	Available	P_A1 P_A2
Available	Reduced	P_A1 P_R2
Available	Unavailable	P_A1 P_U2
Reduced	Available	P_R1 P_A2
Reduced	Reduced	P_R1 P_R2
Reduced	Unavailable	P_R1 P_U2
Unavailable	Available	P_U1 P_A2
Unavailable	Reduced	P_U1 P_R2
Unavailable	Unavailable	P_U1 P_U2

FIGURE 19.2 Weekly load curves by season.

• LOLE (loss of load expectation)—the expected number of hours per year that the system will have to shed load

• EENS (expected energy not served)—the expected number of megawatt hours per year that a system will not be able to supply

System adequacy assessment assumes that the transmission system is available. This may not always be the case. A classic example is the 1965 blackout, which was initiated by the unexpected loss of a transmission line. To address such events, system security assessment is required.

19.3 System Security Assessment

System security is defined as the ability of a power system to supply all of its loads in the event of one or more contingencies (a contingency is an unexpected event such as a system fault or a component outage). This is divided into two separate areas: static security assessment and dynamic security assessment.

Static security assessment determines whether a power system is able to supply peak demand after one or more pieces of equipment (such as a line or a transformer) are disconnected. The system is tested by removing a piece (or multiple pieces) of equipment from the normal power flow model, rerunning the power flow, and determining if all bus voltages are acceptable and all pieces of equipment are loaded below emergency ratings. If an unacceptable voltage or overload violation occurs, load must be shed for this condition and the system is insecure. If removing any single component will not result in the loss of load, the system is N − 1 Secure. If removing any X arbitrary components will not result in the loss of load, the system is N − X Secure. N refers to the number of components on the system and X refers to the number of components that can be safely removed.

Performing a static security assessment can be computationally intensive. For example, an N − 2 assessment on a modest system with 5000 components (1500 buses, 500 transformers, and 3000 lines) will require more than 25 million power flows to be performed. For this reason, contingency ranking methods are often used. These methods rank each contingency based on its likelihood of resulting in load curtailment. Contingencies are examined in order of their contingency ranking, starting with the most severe. If a prespecified number of contingencies are tested and found to be secure, it is assumed that contingencies with less severe rankings are also secure and do not need to be examined.

Static security assessment is based on steady state power flow solutions. For each contingency, it assumes that the system protection has properly operated and the system has reached a steady state. In fact, the power system may not actually reach a steady state after it has been disturbed. Checking whether a system will reach a steady state after a fault occurs is referred to as dynamic security assessment (also referred to as transient security assessment).

When a fault occurs, the system is less able to transfer power from synchronous generators to synchronous motors. Since the instantaneous power input has not changed, generators will begin to speed up and motors will begin to slow down (analogous to the chain slipping while riding a bicycle). This increases the rotor angle difference between generators and motors. If this rotor angle exceeds a critical value, the system will become unstable and the machines will not be able to regain synchronism. After the protection system clears the fault, the rotor angle difference will still increase since the power transfer limits of the system are still less than the prefault condition. If the fault is cleared quickly enough, this additional increase will not cause the rotor angle difference to exceed the critical angle and the system will return to a synchronous state (ABB Power, 1997).

An example of a transient stability test is shown in Figure 19.3. This shows the rotor angle difference between a synchronous generator and a synchronous motor during a fault sequence. When the fault occurs, the rotor angle begins to increase. If the fault is not cleared, the rotor angle quickly exceeds the critical angle. If the fault is cleared at 0.3 s, the rotor angle still increases beyond the critical value. The system is dynamically stable for this fault if it is cleared in 0.2 s. The rotor angle will still increase after the fault occurs but will stabilize below the critical value.

FIGURE 19.3 Dynamic security assessment.

A dynamic security assessment will consist of many transient stability tests that span a broad range of loading conditions, fault locations, and fault types. To reduce the number of tests required, contingency rankings (similar to static security assessment) can be used.

19.4 Probabilistic Security Assessment

Although the “N − 1 Criterion” remains popular, it has received much criticism since it treats unlikely events with the same importance as more frequent events. Using the N − 1 Criterion, large amounts of money may be spent to reinforce a system against a very rare event. From a reliability perspective, this money might be better spent in other areas such as replacing old equipment, decreasing maintenance intervals, adding automation equipment, adding crews, and so on. To make such value judgments, both the impact of each contingency and its probability of occurrence must be considered (Endrenyi, 1978). This is referred to as probabilistic security assessment. To do this type of assessment, each piece of equipment needs at least two fundamental pieces of information: the failure rate of the equipment (usually denoted λ, in failures per year) and the mean time to repair of the equipment (usually denoted MTTR, in hours).

Performing a probabilistic security assessment is similar to a performing a standard static security assessment. First, contingencies are ranked and simulated using a power flow. If a contingency results in the loss of load, information about the number and size of interrupted loads, the frequency of the contingency, and the repair time of the contingency is recorded. This allows quantities such as EENS to be easily computed. If contingency i causes kW_i amount of kilowatts to be interrupted, then EENS is equal to

$E E N S = \sum_{i} k W_{i} λ_{i} M T T R_{i}$ $E E N S = \sum_{i} k W_{i} λ_{i} M T T R_{i}$

(19.2)

It is important to note that this is the EENS due to contingencies and is separate from the EENS due to generation unavailability. It is also important to note that this formula assumes that λ_iMTTR_i is small when compared to 1 year. If this is not the case, a component will experience fewer failures per year than its failure rate and the equation must be adjusted accordingly.

19.5 Distribution System Reliability

The majority of customer reliability problems stem from distribution systems. For a typical residential customer with 90 min of interrupted power per year, between 70 and 80 min will be attributable to problems occurring on the distribution system that it is connected to (Billinton and Jonnavitihula, 1996). This is largely due to radial nature of most distribution systems, the large number of components involved, the sparsity of protection devices and sectionalizing switches, and the proximity of the distribution system to end-use customers.

Since reliability means different things to different people, it is necessary to address the definition of “distribution system reliability” in more detail. In distribution systems, reliability primarily relates to equipment outages and customer interruptions:

• Outage—when a piece of equipment is deenergized

• Momentary interruption—when a customer is deenergized for less than a few minutes

• Sustained interruption—when a customer is deenergized for more than a few minutes

Customers do not, in the strictest sense, experience power outages. Customers experience power interruptions. If power is restored within a few minutes, it is considered a momentary interruption. If not, it is considered a sustained interruption. The precise meaning of “a few minutes” varies from utility to utility, but is typically between 1 and 5 min. The IEEE defines a momentary interruption based on 5 min. (Note: some references classify interruptions into four categories rather than two. Instantaneous interruptions last a few seconds, momentary interruptions last a few minutes, temporary interruptions last a few hours, and sustained interruptions last many hours.)

On a historical note, momentary interruptions used to be considered a “power quality issue” rather than a “reliability issue.” It is now generally agreed that momentary interruptions are an aspect of reliability since (1) momentary interruptions can cause substantial problems to all types of customers, and (2) many trade-offs must be made between momentary interruptions and sustained interruptions during system planning, operation, and control. It can also be observed that customer voltage sags, typically considered a power quality issue, are slowly becoming a reliability issue for similar reasons.

Distribution system reliability is not dependent solely upon component failure characteristics. It is also dependent upon how the system responds to component failures. To understand this, it is necessary to understand the sequence of events that occurs after a distribution system fault.

19.6 Typical Sequence of Events after an Overhead Distribution Fault

The following is a typical sequence of events that will occur after a fault occurs on a distribution system (Brown, 2003).

1. The fault causes high currents to flow from the source to the fault location. These high currents may result in voltage sags for certain customers. These sags can occur on all feeders that have a common coupling at the distribution substation.

2. An instantaneous relay trips open the feeder circuit breaker at the substation. This causes the entire feeder to be deenergized. A pause allows the air around the fault to deionize, and then a reclosing relay will close the circuit breaker. If no fault current is detected, the fault has cleared itself and all customers on the feeder have experienced a momentary interruption.

3. If the fault persists, time overcurrent protection devices are allowed to clear the fault. If the fault is on a fused lateral, the fuse will blow and customers on the lateral will be interrupted. If the feeder breaker trips again, the reclosing relay will repeat the reclosing process a preset number of times before locking out. After the feeder breaker locks out, all customers on the feeder will be interrupted. Automated line switching and system reconfiguration will occur at this point if these capabilities exist.

4. The electric utility will receive trouble calls from customers with interrupted power. It will dispatch a crew to locate the fault and isolate it by opening up surrounding sectionalizing switches. It may also attempt to reconfigure the distribution system in an attempt to restore power to as many customers as possible while the fault is being repaired. Fault isolation can be very fast if switches are motor operated and remotely controlled, but switching usually takes between 15 and 60 min.

5. The crew repairs the faulted equipment and returns the distribution system to its normal operating state.

As can be seen, a fault on the distribution system will impact many different customers in many different ways. In general, the same fault will result in voltage sags for some customers, momentary interruptions for other customers, and varying lengths of sustained interruptions for other customers, depending on how the system is switched and how long the fault takes to repair.

Distribution system reliability assessment methods are able to predict distribution system reliability based on system configuration, system operation, and component reliability data (Brown et al., 1996). This ability is becoming increasingly important as the electric industry becomes more competitive, as regulatory agencies begin to regulate reliability, and as customers begin to demand performance guarantees. The most common reliability assessment methods utilize the following process: (1) they simulate a system’s response to a contingency, (2) they compute the reliability impact that this contingency has on each customer, (3) the reliability impact is weighted by the probability of the contingency occurring, and (4) steps 1–3 are repeated for all contingencies. Since this process results in the reliability that each customer can expect, new designs can be compared, existing systems can be analyzed, and reliability improvement options can be explored.

19.7 Distribution Reliability Indices

Utilities typically keep track of customer reliability by using reliability indices. These are average customer reliability values for a specific area. This area can be the utility’s entire service area, a particular geographic region, a substation service area, a feeder service area, and so on. The most commonly used reliability indices give each customer equal weight. This means that a large industrial customer and a small residential customer will each have an equal impact on computed indices. The most common of these customer reliability indices are as follows: System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration Index (SAIDI), Customer Average Interruption Duration Index (CAIDI), and Average System Availability Index (ASAI) (IEEE, 2003). Notice that these indices are redundant. If SAIFI and SAIDI are known, both CAIDI and ASAI can be calculated. Formulae for these indices are as follows:

$S A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} per year$ $S A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} per year$

(19.3)

$S A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} hours per year$ $S A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} hours per year$

(19.4)

$C A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s} = \frac{S A I D I}{S A I F I} hours per interruption$ $C A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s} = \frac{S A I D I}{S A I F I} hours per interruption$

(19.5)

$A S A I = \frac{C u s t o m e r h o u r s s e r v i c e a v a i l a b i l i t y}{C u s t o m e r h o u r s s e r v i c e d e m a n d} = \frac{8760 - S A I D I}{8760} per unit$ $A S A I = \frac{C u s t o m e r h o u r s s e r v i c e a v a i l a b i l i t y}{C u s t o m e r h o u r s s e r v i c e d e m a n d} = \frac{8760 - S A I D I}{8760} per unit$

(19.6)

Some less commonly used reliability indices are not based on the total number of customers served. The Customer Average Interruption Frequency Index (CAIFI) and the Customer Total Average Interruption Duration Index (CTAIDI) are based upon the number of customers that have experienced one or more interruptions in the relevant year. The Average System Interruption Frequency Index (ASIFI) and the Average System Interruption Duration Index (ASIDI) are based upon the connected kVA of customers (these are sometimes referred to as load-based indices). Formulae for these indices are as follows:

$C A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s}{C u s t o m e r s e x p e r i e n c i n g o n e o r m o r e i n t e r r u p t i o n s} per year$ $C A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r i n t e r r u p t i o n s}{C u s t o m e r s e x p e r i e n c i n g o n e o r m o r e i n t e r r u p t i o n s} per year$

(19.7)

$C T A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{C u s t o m e r s e x p e r i e n c i n g o n e o r m o r e i n t e r r u p t i o n s} hours per year$ $C T A I D I = \frac{\sum C u s t o m e r i n t e r r u p t i o n d u r a t i o n s}{C u s t o m e r s e x p e r i e n c i n g o n e o r m o r e i n t e r r u p t i o n s} hours per year$

(19.8)

$A S I F I = \frac{C o n n e c t e d k V A i n t e r r u p t e d}{T o t a l c o n n e c t e d k V A s e r v e d} per year$ $A S I F I = \frac{C o n n e c t e d k V A i n t e r r u p t e d}{T o t a l c o n n e c t e d k V A s e r v e d} per year$

(19.9)

$A S I D I = \frac{C o n n e c t e d k V A h o u r s i n t e r r u p t e d}{T o t a l c o n n e c t e d k V A s e r v e d} hours per year$ $A S I D I = \frac{C o n n e c t e d k V A h o u r s i n t e r r u p t e d}{T o t a l c o n n e c t e d k V A s e r v e d} hours per year$

(19.10)

As momentary interruptions become more important, it becomes necessary to keep track of indices related to momentary interruptions. Since the duration of momentary interruptions is of little consequence, a single frequency-related index, the Momentary Average Interruption Frequency Index (MAIFI), is all that is needed. MAIFI, like SAIFI, weights each customer equally (there is currently no load-based index for momentary interruptions). The formula for MAIFI is as follows:

$M A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r m o m e n t a r y i n t e r r u p t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} per year$ $M A I F I = \frac{T o t a l n u m b e r o f c u s t o m e r m o m e n t a r y i n t e r r u p t i o n s}{T o t a l n u m b e r o f c u s t o m e r s s e r v e d} per year$

(19.11)

The precise application of MAIFI varies. This variation is best illustrated by an example. Assume that a customer experiences three recloser operations followed by a recloser lockout, all within a period of 1 min. Some utilities would not count this event as a momentary interruption since the customer experiences a sustained interruption. Other utilities would count this event as three momentary interruptions and one sustained interruption. Similarly, if a customer experiences three recloser operations within a period of 1 min with power being restored after the last recloser, some utilities would count the event as three momentary interruptions and other utilities would count the event as a single momentary interruption. The IEEE defines a momentary interruption as lasting less than five minutes and a “momentary event” as a grouping of one or more momentary interruptions occurring within a five minute interval.

19.8 Storms and Major Events

When electric utilities compute reliability indices, they often exclude interruptions caused by “storms” and “major events.” The definition of a major event varies from utility to utility, but a typical example is when more than 10% of customers experience an interruption during the event. The event starts when the notification of the first interruption is received and ends when all customers are restored service. The IEEE defines a major event based on a statistical approach called 2.5 Beta (IEEE, 2003).

In nonstorm conditions, equipment failures are independent events—the failure of one device is completely independent of another device. In contrast, major events are characterized by common-mode failures. This means that a common cause is responsible for all equipment failures. The result is that many components tend to fail at the same time. This puts a strain on utility resources, which can only handle a certain number of concurrent failures (Brown et al., 1997). The most common causes of major events are wind storms, ice storms, and heat waves.

Wind storms refer to linear winds that blow down trees and utility poles. The severity of wind storms is dependent upon sustained wind speed, gust speed, wind direction, and the length of the storm. Severity is also sensitive to vegetation management and the time elapsed since the last wind storm. Since a wind storm will tend to blow over all of the weak trees, a similar storm occurring a few months later may have little impact. A U.S. map showing wind speeds for the worst expected storm in 50 years is shown in Figure 19.4.

FIGURE 19.4 Fifty-year wind storm (sustained wind speed in miles/h).

Ice storms refer to ice buildup on conductors. This has four major effects: (1) it places a heavy physical load on the conductors and support structures, (2) it increases the cross-sectional area that is exposed to the wind, (3) ice can break off and cause a conductor to jump into the phase wires located above it, and (4) galloping. Galloping occurs when ice buildup assumes a teardrop shape and acts as an airfoil. During high winds, this can cause conductors to swing wildly and with great force. Ice can also cause problems by accumulating in trees, causing limbs to break off, and causing entire trunks to fall over into power lines.

Heat waves are extended periods of exceedingly hot weather. This hot weather causes electricity demand to skyrocket due to air-conditioning loads. At the same time, conductors cannot carry as much electricity since they cannot transfer heat as effectively to their surroundings. This combination of heavy loading and conductor de-rating can cause overhead wires to become overloaded and sag to dangerous levels. Overloaded cables will cause insulation to lose life. In a worst-case scenario, the maximum power transfer capabilities of the system can be approached, resulting in a voltage collapse condition. Humidity exacerbates the impact of heat waves since it causes air conditioners to consume more energy.

19.9 Component Reliability Data

For a reliability model to be accurate, component reliability data must be representative of the system being modeled. Utilities recognize this and are increasing their efforts to keep track of component failure rates, failure modes, repair times, switching times, and other important reliability parameters. Unfortunately, reliability statistics vary widely from utility to utility and from country to country. The range of equipment reliability data that can be found in published literature is shown in Table 19.2.

Because component reliability is very system specific, it is beneficial to calibrate reliability models to historical reliability indices. In this process, component reliability parameters are adjusted until historical reliability indices match computed reliability indices (Brown and Ochoa, 1998). The amount that each parameter is adjusted should depend on the confidence of the original value and the sensitivity of the reliability indices to changes in this value. To illustrate, consider an overhead distribution system. A reliability model of this system is created using component reliability data from published literature. Unfortunately, the reliability indices that the model produces do not agree with the historical performance of the system over the past few years. To fix this, the failure rate and repair times of overhead lines (along with other component parameters) can be adjusted until predicted reliability matches historical reliability.

TABLE 19.2 Equipment Reliability Data

Component	Failure Rate (year⁻¹)	MTTR (h)
Substation Equipment
Power transformers	0.015–0.07	15–480
Circuit breakers	0.003–0.02	6–80
Disconnect switches	0.004–0.16	1.5–12
Air-insulated buswork	0.002–0.04	2–13
Overhead Equipment
Transmission lines^a	0.003–0.140	4–280
Distribution lines^a	0.030–0.180	4–110
Switches/fused cutouts	0.004–0.014	1–4
Pole mounted transformer	0.001–0.004	3–8
Underground Equipment
Cable^a	0.005–0.04	3–30
Padmount switches	0.001–0.01	1–5
Padmount transformers	0.002–0.003	2–6
Cable terminations/joints	0.0001–0.002	2–4

^a Failure rates for lines and cable are per mile.

19.10 Utility Reliability Problems

To gain a broader understanding of power system reliability, it is necessary to understand the root causes of system faults and system failures. A description of major failure modes is now provided.

19.10.1 Underground Cable

A major reliability concern pertaining to underground cables is electrochemical treeing. Treeing occurs when moisture penetration in the presence of an electric field reduces the dielectric strength of cable insulation. When the dielectric strength is degraded sufficiently, transients caused by lightning or switching can result in dielectric breakdown. Electrochemical treeing usually affects extruded dielectric cable such as cross-linked polyethylene (XLPE) and ethylene-propylene rubber (EPR), and is largely attributed to insulation impurities and bad manufacturing. To reduce failures related to electrochemical treeing, a utility can install surge protection on riser poles (transitions from overhead to underground), can purchase tree-retardant cable, and can test cable reels before accepting them from the manufacturer.

Existing cable can be tested and replaced if problems are found. One way to do this is to apply a DC voltage withstand test (approximately three times nominal RMS voltage). Since cables will either pass or not pass this test, information about the state of cable deterioration cannot be determined. Another popular method for cable testing is to inject a small signal into one end and check for reflections that will occur at partial discharge points. Other methods are measuring the power factor over a range of frequencies (dielectric spectroscopy), analyzing physical insulation samples in a lab for polymeric breakdown (degree of polymerization), and using cable indentors to test the hardness of the insulation.

Not all underground cable system failures are due to cable insulation. A substantial percentage occurs at splices, terminations, and joints. Major causes are due to water ingress and poor workmanship. Heat shrink covers can be used to waterproof these junctions and improve reliability.

The last major reliability concern for underground cable is dig-ins. This is when excavation equipment cuts through one or more cables. To prevent dig-ins, utilities should encourage the public to have cable routes identified before initiating site excavation. In extreme cases where high reliability is required, utilities can place cable in concrete-encased duct banks.

19.10.2 Transformer Failures

Transformers are critical links in power systems, and can take a long time to replace if they fail. Through-faults cause extreme physical stress on transformer windings, and are the major cause of transformer failures. Overloads rarely result in transformer failures, but do cause thermal aging of winding insulation.

When a transformer becomes hot, the insulation on the windings slowly breaks down and becomes brittle over time. The rate of thermal breakdown approximately doubles for every 10°C. Ten degree Celcius is referred to as the “Montsinger Factor” and is a rule of thumb describing the Arrhenius theory of electrolytic dissociation. Because of this exponential relationship, transformer overloads can result in rapid transformer aging. When thermal aging has caused insulation to become sufficiently brittle, the next fault current that passes through the transformer will mechanically shake the windings, a crack will form in the insulation, and an internal transformer fault will result.

Extreme hot-spot temperatures in liquid-filled transformers can also result in failure. This is because the hot spot can cause free bubbles that reduce the dielectric strength of the liquid. Even if free bubbles are not formed, high temperatures will increase internal tank pressure and may result in overflow or tank rupture.

Many transformers are fitted with load tap changers (LTCs) for voltage regulation. These mechanically moving devices have historically been prone to failure and can substantially reduce the reliability of a transformer (Willis, 1997). Manufacturers have addressed this problem and new LTC models using vacuum technology have succeeded in reducing failure rates.

19.10.3 Lightning

A lightning strike occurs when the voltage generated between a cloud and the ground exceeds the dielectric strength of the air. This results in a massive current stroke that usually exceeds 30,000 A. To make matters worse, most strokes consist of multiple discharges within a fraction of a second. Lightning is the major reliability concern for utilities located in high keraunic areas (Burke, 1994). An isokeraunic map for the United States is shown in Figure 19.5.

FIGURE 19.5 Number of thunderstorm days per year.

Lightning can affect power systems through direct strikes (the stroke contacts the power system) or through indirect strikes (the stroke contacts something in close proximity and induces a traveling voltage wave on the power system). Lightning can be protected against by having a high system BIL (basic impulse level) by using shield wires, by using surge arrestors to clamp voltages across equipment, and by having a low impedance ground. Direct strikes are virtually impossible to protect against on a distribution system.

19.10.4 Tree Contact

Trees continuously grow, can fall over onto conductors, can drop branches onto conductors, can push conductors together, and can serve as gateway for animals. This is why many utilities spend more on tree trimming than on any other preventative maintenance activity.

When a tree branch bridges two conductors, a fault does not occur immediately. This is because a moist tree branch has a substantial resistance. A small current begins to flow and starts to dry out the wood fibers. After several minutes, the cellulose will carbonize, resistance will be greatly reduced, and a short circuit will occur. Branches brushing against a single phase conductor typically do not result in system faults.

Faults due to tree contact can be reduced by using tree wire. This is overhead wire with an insulated jacket similar to cable. Tree wire can be effective, but faults tend to result in conductor burndown since they will not motor (move themselves along the conductor) like faults on bare conductor.

19.10.5 Birds

Birds are the most common cause of animal faults on both transmission systems and air-insulated substations. Different types of birds cause different types of problems, but they can generally be classified as nesting birds, roosting birds, raptors, and woodpeckers.

Nesting birds commonly build their homes on transmission towers and in substations. Nesting materials can cause faults, and bird excrement can contaminate insulators. Nesting birds also attract predators such as raccoons, snakes, and cats. These predators can be a worse reliability problem than the birds themselves.

Roosting birds use electrical equipment to rest on or to search for prey. They can be electrocuted by bridging conductors with their wings, and their excrement can contaminate insulators. To prevent birds from roosting, anti-roosting devices can be placed on attractive sites. For locations that cater to thousands of roosting birds, more extreme deterrent methods such as pyrotechnics can be used.

Raptors are birds of prey such as eagles, hawks, ospreys, owls, and vultures. Reliability problems are similar to other roosting and nesting birds, but special consideration may be required since most raptors are protected by the federal government.

Woodpeckers peck holes in wood with their beaks as they search for insects. This does not harm trees (the bark regenerates), but can cause devastating damage to utility poles. This can be prevented by using steel poles, by using repellent, or by tricking a woodpecker into believing that there is already a resident woodpecker (woodpeckers are quite territorial).

19.10.6 Squirrels

Squirrels are a reliability concern for all overhead distribution systems near wooded areas. Squirrels will not typically climb utility poles, but will leap onto them from nearby trees. They cause faults by bridging grounded equipment with phase conductors. Squirrel problems can be mitigated by cutting down nearby access trees or by installing animal guards on insulators.

19.10.7 Snakes

Snakes are major reliability concerns in both substations and underground systems. They can squeeze through very small openings, can climb almost anything, and have the length to easily span phase conductors. Snakes are usually searching for food (birds in substations and mice in underground systems), and removing the food supply can often remove the snake problem. Special “snake fences” are also available.

19.10.8 Insects

It is becoming more common for fire ants to build nests in pad-mounted equipment. Their nesting materials can cause short circuits, the ants can eat away at conductor insulation, and they make equipment maintenance a challenge.

19.10.9 Bears, Bison, and Cattle

These large animals do not typically cause short circuits, but degrade the structural integrity of poles by rubbing on guy wires. Bears can also destroy wooden poles by using them as scratching posts, and black bears can climb wooden utility poles. These problems can be addressed by placing fences around poles and guy wire anchors.

19.10.10 Mice, Rats, and Gophers

These rodents cause faults by gnawing through the insulation of underground cable. They are the most common cause of animal-related outages on underground equipment. To make matters worse, they will attract snakes (also a reliability problem). Equipment cabinets should be tightly sealed to prevent these small animals from entering. Ultrasonic devices can also be used to keep rodents away (ultrasonic devices will not keep snakes away).

19.10.11 Vandalism

Vandalism can take many different forms, from people shooting insulators with rifles to professional thieves stealing conductor wire for scrap metal. Addressing these reliability problems will vary greatly from situation to situation.

19.11 Reliability Economics

When a power interruption occurs, both the utility and the interrupted customers are inconvenienced. The utility must spend money to fix the problem, will lose energy sales during the interruption, and may be sued by disgruntled customers. From the customer perspective, batch processes may be ruined, electronic devices may crash, production may be lost, retail sales may be lost, and inventory (such as refrigerated food) may be ruined.

When a customer experiences an interruption, there is an amount of money that it would be willing to pay to have avoided the interruption. This amount is referred to as the customer’s incurred cost of poor reliability, and consists of a base cost plus a time-dependent cost. The base cost is the same for all interruptions, relates to electronic equipment shutdown and interrupted processes, and is equivalent to the cost of a momentary interruption. The time-dependent cost relates to lost production and extended inconvenience, and reflects that customers would prefer interruptions to be shorter rather than longer.

The customer cost of an interruption varies widely from customer to customer and from country to country. Other important factors include the time of year, the day of the week, the time of day, and whether advanced warning is provided. Specific results are well documented by a host of customer surveys (Billinton et al., 1983; IEEE Std. 493-1990; Tollefson et al., 1991, 1994). For planning purposes, it is useful to aggregate these results into a few basic customer classes: commercial, industrial, and residential. Since larger customers will have a higher cost of reliability, results are normalized to the peak kW load of each customer. Reliability cost curves for typical U.S. customers are shown in Figure 19.6.

FIGURE 19.6 Typical U.S. customer interruption costs (1999 dollars).

Average customer cost curves tend to be linear and can be modeled as an initial cost plus a first order, time-dependent cost. Specific customer cost curves may be extremely nonlinear. For example, a meat packing warehouse depending upon refrigeration may be unaffected by interruptions lasting many hours. At a certain point, the meat will begin to spoil and severe economic losses will quickly occur. After the meat spoils, additional interruption time will harm this particular customer much more.

19.12 Annual Variations in Reliability

Power system reliability varies from year to year. In a lucky year, a system may have a SAIDI of 30 min. The next year, this exact same system may experience a SAIDI of 8 h. This type of variation is inevitable and must be considered when comparing reliability indices. It is also important to note that the variance of reliability indices will tend to be less for areas serving more customers. Individual customer reliability will tend to be the most volatile, followed by feeder reliability, substation reliability, regional reliability, and so forth.

The importance of annual reliability variance will grow as utilities become subject to performance-based rates and as customer reliability guarantees become more common. These types of contracts expose utilities to new risks that must be understood and managed. Since performance-based contracts penalize and reward utilities based on reliability, annual variations must be understood for fair contracts to be negotiated and managed.

Contractual issues concerning service reliability are becoming important as the electric industry becomes more competitive. Customers can choose between suppliers, and realize that there is a tradeoff between reliability and rates. Some customers will demand poor reliability at low rates, and other customers will demand high reliability at premium rates. To address the wide variation in customer needs, utilities can no longer be suppliers of energy alone, but must become suppliers of both energy and reliability. Power system reliability is now a bona fide commodity with explicit value for utilities to supply and explicit value that customers demand.

References

Allan, R.N., Billinton, R., Breipohl, A.M., and Grigg, C.H., Bibliography on the application of probability methods in power system reliability evaluation, IEEE Trans. Power Syst., 9, 1, 41–49, Feb. 1994.

Billinton, R. and Allan, R.N., Reliability Assessment of Large Electric Power Systems, Kluwer Academic Publishers, Dordrecht, the Netherlands, 1988.

Billinton, R. and Allan, R., Reliability Evaluation of Engineering Systems: Concepts and Techniques, 2nd edn., Plenum Press, New York, 1992.

Billinton, R. and Jonnavitihula, S., A test system for teaching overall power system reliability assessment, IEEE Trans. Power Syst., 11, 4, 1670–1676, Nov. 1996.

Billinton, R., Wacker, G., and Wojczynski, E., Comprehensive bibliography on electrical service interruption costs, IEEE Trans. Power Appar. Syst., PAS-102, 6, 1831–1837, June 1983.

Brown, R.E., Electric Power Distribution Reliability, Second Edition, CRC Press, 2009.

Brown, R.E., Gupta, S., Christie, R.D., Venkata, S.S., and Fletcher, R.D., Distribution system reliability analysis using hierarchical Markov modeling, IEEE Trans. Power Delivery, 11, 4, 1929–1934, Oct. 1996.

Brown, R.E., Gupta, S., Christie, R.D., Venkata, S.S., and Fletcher, R.D., Distribution system reliability: Momentary interruptions and storms, IEEE Trans. Power Delivery, 12, 4, 1569–1575, Oct. 1997.

Brown, R.E. and Ochoa, J.R., Distribution system reliability: Default data and model validation, IEEE Trans. Power Syst., 13, 2, 704–709, May 1998.

Burke, J.J., Power Distribution Engineering, Marcel Dekker, Inc., New York, 1994.

Electrical Transmission and Distribution Reference Book, ABB Power T&D Company, Inc., Raleigh, NC, 1997.

Endrenyi, J., Reliability in Electric Power Systems, John Wiley & Sons, Ltd., New York, 1978.

IEEE Std. 493-1990, IEEE Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems, 1990.

IEEE Standard 1366–2003, Guide for Electric Power Distribution Reliability Indices, 2003.

Philipson, L. and Willis, H.L., Understanding Electric Utilities and De-regulation, Marcel Dekker, Inc., New York, 1999.

Ramakumar, R., Engineering Reliability: Fundamentals and Applications, Prentice-Hall, Inc., Englewood Cliffs, NJ, 1993.

Schilling, M.T., Billinton, R., Leite da Silva, A.M., and El-Kady, M.A., Bibliography on composite system reliability (1964–1988), IEEE Trans. Power Syst., 4, 3, 1122–1132, Aug. 1989.

Tollefson, G., Billinton, R., and Wacker, G., Comprehensive bibliography on reliability worth and electric service consumer interruption costs 1980–1990, IEEE Trans. Power Syst., 6, 4, 1508–1514, Nov. 1991.

Tollefson, G., Billinton, R., Wacker, G., Chan, E., and Aweya, J., A Canadian customer survey to assess power system reliability worth, IEEE Trans. Power Syst., 9, 1, 443–450, Feb. 1994.

Willis, H.L., Power Distribution Planning Reference Book, Marcel Dekker, Inc., New York, 1997.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for 19 Power System Reliability

Create new playlist

Sign In

Sign Up

Table of Contents for
19 Power System Reliability